This paper exposes problems of the commonly used technique of splitting the available data in neural spatial interaction modelling into training, validation, and test sets that are held fixed and warns about drawing too strong conclusions from such static splits. Using a bootstrapping procedure, we compare the uncertainty in the solution stemming from the data splitting with model specific uncertainties such as parameter initialization. Utilizing the Austrian interregional telecommunication traffic data and the differential evolution method for solving the parameter estimation task for a fixed topology of the network model [ i.e. J = 9] this paper illustrates that the variation due to different resamplings is significantly larger than the variation due to different parameter initializations. This result implies that it is important to not over-interpret a model, estimated on one specific static split of the data. (authors' abstract) / Series: Discussion Papers of the Institute for Economic Geography and GIScience

Predicting the passenger flow inside a city is a vital component of the intelligent transportation management system. The proposal for a new residential area, an office space, post­pandemic policy implications for work from home, behavioral changes for revised traffic patterns, infrastructural improvements, require a visual and analytical backing which can be provided through a macro simulation model. This research explores the performance of the Machine learning (ML) based transport model against the predictions provided by the traditional Spatial Interaction Models (SIM) for the city of Oslo. The transport models and their parameters are analyzed for sensitivity analysis and scenario analysis to derive city character. Furthermore, the derived model is deployed over an interactive dashboard for analytical and their practical visualizations through infographics. The results show that the ML model outperforms the SIM. Although the traditional SIM has a clear advantage of being interpreted by design and requiring a few parameters, it suffers from its inability to accurately capture the structure of real flows and greater variability as compared to the ML model. Extensive statistical analyses are conducted to obtain significant results and realize the pros and cons of both the models which question the validity of results for the ML model over SIM. With this thesis, we discuss the potential of ML model detected trends of passenger flows, andtheir capacity to simulate city development­related scenarios for the traffic flows within the city.